Talk:Last letter-first letter: Difference between revisions
→Full List of Names, and Input to Tasks in General
| (7 intermediate revisions by 5 users not shown) | |||
Line 6:
Re problems with PicoLisp version, Nidoran can be male or female, and this is marked in the list with male and female (mars and venus) symbols. I'd suggest spelling out "Male" and "Female". [[User:Axtens|Axtens]] 10:10, 7 June 2011 (UTC)
: OK, the
:: I complained because it flatly wasn't obeying the rules of the game. The rules relate to letters of words, not random extra obscure marks. Right is right! –[[User:Dkf|Donal Fellows]] 14:45, 7 June 2011 (UTC)
Line 67:
::+1 :[[User:Sonia|Sonia]] 21:27, 8 June 2011 (UTC)
===Full List of Names, and Input to Tasks in General===
I have included my list of the 646 names at the bottom of the Racket implementation; it clutters up the implementation (and the page as a whole) in a manner that I feel is undesirable.
Is what I have done good form -- because on the other hand, people can use the code I have submitted, and they can use my list as a basis of their own.
* Can I "attach" the file to the page and refer to it via a link, so it doesn't have to appear?
* Can I put a "fold" in the markup?
* Should I host the file elsewhere?
Also, is there any way to "semantically" tag the list of 70 names in the task (or any of the sample input that is specified for tasks, in general) so they can be automatically scraped by implementations? --[[User:Tim-brown|tim-brown]] ([[User talk:Tim-brown|talk]]) 07:40, 22 May 2013 (UTC)
: Where we have a particularly long entry for a specific language, we often put that entry (or the bulky part of it) on another page. For example, here, you might use Last_letter-first_letter/Racket. And, here, since the full list of names was not a part of the task, I think it makes sense to leave the core entry in place and just move the "extras".
: But this is not the only possibility (for example, in some cases it makes sense to refer to an external site for "going further").
: As for the "raw data" aspect - hypothetically speaking, that should be addressed in the task description (other than that... it's just like any other issue that leads to bulk, so of course if your language context suggests something specific be said about the data you can express that). --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 12:58, 22 May 2013 (UTC)
== Go "two interpretations of longest" ==
Line 85 ⟶ 102:
::: Do you want the removal of all timings then?
:::: Some timings can be worthwhile. For example, when you have several different timings based on the same platform, and time differences greater than a factor of 2, those timings might be meaningful. --[[User:Rdm|Rdm]] 03:54, 29 June 2011 (UTC)
:::: I would strongly advise against comparing timings of languages on Rosetta Code, except when comparing performance of different means of using the same language ''on the same hardware.'' Otherwise, I don't see the conditions Rosetta Code offers to be controlled, stable and consistent enough to result in the display of good, useful comparative data, and would far more likely result in destructive, uninformed comparisons used in advocacy by attack. That kind of data is what the Computer Language Shootout discusses. If, for some reason, their numbers are unsatisfactory, talk to them. If they disagree with you, talk to me, and I might talk to them. I have no illusion that I'd be able to convince them, but it's plausible there are some problems I could help address. I don't want RC being used as any kind of an authoritative source for something where we can't maintain the kind of QC necessary. --[[User:Short Circuit|Michael Mol]] 04:29, 29 June 2011 (UTC)
::::You really want to leave timings out except for the most general of comparitive terms, without the use of numbers and without changes in language or run environment is preferable to stop "miss-use".--
:::::Correct. If language-language speed comparisons of particular problems is truly desired, I'd need to set up some kind of controlled conditions. That would be painfully difficult to do properly and securely, considering competing interests and need to get language-specific tuning parameters correct. --[[User:Short Circuit|Michael Mol]] 12:57, 29 June 2011 (UTC)
::::::Also, the optimizations needed to get good speed (sometimes even just decent speed) out of certain tasks would make them hard to read, which IIRC was an important consideration. [[User:MagiMaster|MagiMaster]] 00:36, 30 June 2011 (UTC)
:::::::From time to time, I've seen people post "This method works, but is inefficient. (code example) A more efficient method would be to (narrative describing transform of code), like this: (alternate code example)." --[[User:Short Circuit|Michael Mol]] 13:50, 30 June 2011 (UTC)
== Alternative methods? ==
I am toying with an annealing kind of code to make longest chains, and it was able to come up with a 241 sequence in a minute or so (645 names). For reference, for the 70 name sequence it was able to find a 21. If the full list can't be brute-forced, maybe some approximate methods would be interesting. --[[User:Ledrug|Ledrug]] 02:22, 30 June 2011 (UTC)
| |||